Canonical Quantization On The Half-Line and in An Interval Based Upon A New Concept For The Momentum in A Space With Boundaries

Canonical quantization on the half-line and in an interval based upon
a new concept for the momentum in a space with boundaries

M. H. Al-Hashimi and U.-J. Wiese†
†
Albert Einstein Center, Institute for Theoretical Physics, University of Bern, 3012 Bern, Switzerland
For a particle moving on a half-line or in an interval the operator p̂ = −i∂x is not self-adjoint
and thus does not qualify as the physical momentum. Consequently canonical quantization based
on p̂ fails. Based upon a new concept for a self-adjoint momentum operator p̂R , we show that
canonical quantization can indeed be implemented on the half-line and on an interval. Both the
Hamiltonian Ĥ and the momentum operator p̂R are endowed with self-adjoint extension parameters
that characterize the corresponding domains D(Ĥ) and D(p̂R ) in the Hilbert space. When one
replaces Poisson brackets by commutators, one obtains meaningful results only if the corresponding
operator domains are properly taken into account. The new concept for the momentum is used to
arXiv:2103.01715v1 [quant-ph] 2 Mar 2021
describe the results of momentum measurements of a quantum mechanical particle that is reflected
at impenetrable boundaries, either at the end of the half-line or at the two ends of an interval.
I. INTRODUCTION blocks of higher-dimensional regions of space with sharp

boundaries. The present investigations hence form the
Momentum is one of the most fundamental physical basis for future applications to physical systems in higher
quantities. The momentum operator generates infinitesi- dimensions. The sharp boundaries of a confined system
mal translations in infinite space. When the Hamiltonian give rise to a high degree of ultraviolet sensitivity. At low
is translation invariant, momentum is conserved. Coor- energies, this reflects itself in the values of self-adjoint ex-
dinates and their canonically conjugate momenta play a tension parameters of the Hamiltonian that characterize
central role in Hamiltonian dynamics, which turns into the boundary conditions of the wave function.
quantum mechanics upon canonical quantization. In order to take the ultraviolet sensitivity into account
Several important physical systems are confined in- properly, in our construction of a self-adjoint momentum
side a limited region of space with sharp boundaries. operator in [9] we started out from an ultraviolet lattice
These include, for example, ultra-cold atoms in an op- regularization. While this is familiar in non-perturbative
tical box trap [1] of a few tens of micrometers in size, quantum field theory, in this case it is beneficial even
electrons in a quantum dot at the nanometer scale [2], in quantum mechanics. On the lattice, one naturally
the “femto-universe” of the phenomenological MIT bag distinguishes forward and backward derivatives, neither
model [3–5] for confined quarks and gluons, domain wall of them being Hermitean. The Hermitean lattice mo-
fermions [6, 7], or intervals of extra-dimensional space at mentum operator is given by a symmetrized forward-
the Planck scale. In the idealized mathematical descrip- backward derivative, which manifests itself as a finite
tion of such systems, space is endowed with sharp impen- difference that extends over two lattice spacings. This
etrable boundaries. While for most systems this is just a naturally leads to the distinction of even and odd lattice
mathematical convenience that allows one to exclude re- points. A careful analysis of the problem reveals that the
gions of very high potential energy, it is conceivable that construction of a self-adjoint momentum operator in the
extra-dimensional space literally ends at a boundary. In continuum limit requires the doubling of the standard
spaces with boundaries, translation invariance is explic- Hilbert space, in order to maintain a remnant of the
itly broken, not just because the Hamiltonian includes crucial distinction between even and odd lattice points
symmetry-breaking terms, but because space itself ends even in the continuum [9]. In fact, some aspects of the
(at least in the idealized mathematical description). As a problem are reminiscent of the lattice fermion doubling
consequence, the usual quantum mechanical momentum problem [10–12], which arises because the Dirac operator
operator p̂ = −i∂x is no longer self-adjoint, and thus contains first-order derivatives. The resulting insight of
no longer represents the physical momentum. From this [9] is that the construction of a self-adjoint momentum
it has been concluded that, in a finite volume, momen- operator requires a refined concept, not only of Hilbert
tum is no longer a measurable physical observable [8]. space, but even of space itself. This may not be too
As another consequence of the non-self-adjointness of p̂, surprising, because a space that ends abruptly supports
canonical quantization (which is based upon p̂) fails for other momentum modes than infinite space. Still, as we
confined systems with sharp boundaries. will see, the finite-energy sector, which is defined by the
Recently, we have introduced a new concept for the Hamiltonian, resides in a region of Hilbert space that is
momentum of a quantum mechanical particle in a box completely equivalent to the standard quantum mechan-
[9], which gives rise to a physically and mathematically ical treatment.
satisfactory self-adjoint momentum operator. This con- An extension of the Hilbert space also plays an im-
cept naturally extends to the half-line as well as to higher portant role in quantum measurements related to a pos-
dimensions. Intervals and half-lines are the basic building itive operator-valued measure (POVM) [13–19], which
2
are based on Kraus operators describing quantum jumps. scribes the measured observable. It is possible that the
Measurement processes using a POVM provide a gener- domains of the Hamiltonian D(Ĥ) and of the observ-
alization of von Neumann’s standard projective measure- able D(Â) do not coincide. This is no problem, because
ments. They play an important role for controlling quan- a self-adjoint operator has a complete set of eigenfunc-
tum systems and processing quantum information. In tions. As a consequence, every wave function (even out-
POVM-based measurements, the quantum system to be side of D(Â)) can be represented arbitrarily well by a
investigated is first coupled to another quantum system superposition of eigenstates of Â. In this way, one can
that acts as an ancilla. Then a standard projective mea- determine the probabilities to measure the various pos-
surement is performed on the ancilla, which indirectly af- sible eigenvalues of the observable Â. After such a pro-
fects the quantum system under study [20]. That a given jective measurement, the wave function is inside D(Â)
POVM can be realized by an appropriate extension of the
but not necessarily any longer inside D(Ĥ). How can
Hilbert space is related to Naimark’s theorem [18, 21–24].
the unitary time-evolution proceed after such a measure-
In this way measurements on confined systems have been
ment? Again, since the Hamiltonian is self-adjoint, any
described by POVMs [25] and an optimal POVM for a
state (even outside D(Ĥ)) can be approximated arbitrar-
particle on a half-line has been considered in [26]. It
should be pointed out that the POVM measurement is ily well by a superposition of eigenstates of Ĥ, and the
not based on a self-adjoint momentum operator of the time-evolution proceeds accordingly. However, the en-
quantum particle itself, but rather of the particle cou- ergy expectation value after a measurement that leads
pled to its ancilla. Our construction, on the other hand, out of D(Ĥ) is usually infinite. Hence, idealized mea-
provides a self-adjoint momentum operator for the parti- surements can transfer an infinite amount of energy to
cle alone (without invoking any ancilla). In our case, the the particle under investigation. Of course, any practical
doubling of the Hilbert space results from the necessity measurement only consumes a finite amount of energy,
to include states whose energy is ultraviolet-sensitive, but and is, in any case, not completely realistically described
which, due to the existence of sharp boundaries, still con- by an idealized projective measurement.
tribute to the momentum eigenstates. In other words, the Momentum measurements on confined particles fall in
Hilbert space that contains the finite-energy eigenstates this category. For example, a particle of finite energy
is too small to contain also the momentum eigenstates of that moves along the entire real axis in a potential that
a quantum particle in a space with sharp boundaries. diverges at spatial infinity, V (±∞) → ∞, has a square-
The time-evolution of a quantum system is driven by integrable wave function that vanishes at infinity and
its Hamiltonian, which is described by a self-adjoint op- belongs to the Hilbert space L2 (R) of square-integrable
erator acting in an appropriate Hilbert space. In non- wave functions over the entire real axis. When the mo-
relativistic 1-d quantum mechanics, the single-particle mentum of the particle is measured and one obtains
1
Hamiltonian Ĥ = − 2m ∂x2 + V (x) (in units where ~ = 1) the value k, its wave function collapses onto the plane
contains the differential operator ∂x2 , which is supposed wave momentum eigenstate hx|ki = exp(ikx), which has
to act on square-integrable wave functions. Since not all hk|V |ki = ∞. Hence, an idealized momentum measure-
square-integrable functions are differentiable, the Hamil- ment indeed transfers an infinite amount of energy to a
tonian (as well as other physical operators) act only in confined particle.
a restricted domain D(Ĥ) of the Hilbert space. For a The situation becomes more subtle in the presence of
differential operator the domain is characterized by the impenetrable sharp boundaries (which, of course, again
square-integrability of the corresponding derivatives of are mathematical idealizations) [8, 31–35]. For exam-
the wave function. In an infinite-dimensional Hilbert ple, for a particle that is strictly confined to the positive
space, there are subtle differences between Hermiticity real axis (the half-line) the operator p̂ = −i∂x is not self-
and self-adjointness, which were first understood by von adjoint, and hence it has until now been unknown how to
Neumann [27]. Hermiticity means that an operator Â properly define the corresponding momentum operator.
and its adjoint Â† act in the same way. Self-adjointness As a consequence, one has replaced the momentum op-
requires, in addition, that the corresponding domains erator by the dilation operator, thus moving from canon-
D(Â) = D(Â† ) coincide [28–30]. In order to qualify as ical to affine quantization [36–40]. The main purpose of
a physical observable, an operator must be self-adjoint. this paper is to provide an appropriate construction of a
This is because only self-adjointness, and not Hermitic- self-adjoint momentum operator that is satisfactory both
ity alone, guarantees a spectrum of real eigenvalues with from a physical and from a mathematical point of view,
a corresponding complete set of orthonormal eigenfunc- and to show that canonical quantization is, in fact, appli-
tions. During its time-evolution the wave function of cable to the half-line as well as to an interval. As a result,
a particle with finite energy only explores the domain we will be able to describe momentum measurements per-
D(Ĥ), and never reaches other corners in Hilbert space. formed on a quantum mechanical particle that is strictly
However, when the unitary time-evolution driven by limited to the positive real axis or to an interval, even af-
the Hamiltonian is interrupted by an (idealized) pro- ter a momentum measurement. Such a particle is bound
jective measurement, the momentary wave function is to or reflected at impenetrable boundaries, either at the
projected onto an eigenstate of the operator Â that de- origin or at the two ends of the interval.
3
It should be pointed out that our construction of a self- tization. Based upon our new concept, in Section III
adjoint momentum operator assumes that some physi- we construct a self-adjoint momentum operator on the
cally meaningful ultraviolet cut-off (like a crystal lattice half-line, embed the Hamiltonian in the resulting math-
in a quantum dot) actually exists, and that the quan- ematical framework, and discuss the corresponding mo-
tum mechanical description in the continuum is an ef- mentum measurements. We then consider the resulting
fective low-energy description that is valid only below canonical quantization on the half-line including the clas-
that ultraviolet cut-off. If one assumes that the quantum sical limit. Section IV addresses canonical quantization
mechanics formulated in the continuum is a “theory of on an interval and relates the results to the situation on
everything”, in other words that no physical ultraviolet a circle. Finally, we end with some conclusions. Ultra-
cut-off exists at short distances, there would be no basis violet lattice aspects of the new momentum concept are
for extending the Hilbert space. In that case, momentum summarized in an appendix.
would indeed not be a meaningful concept for motion on
the half-line, and one would be restricted to affine quanti-
zation. In experimental situations involving, for example,
quantum dots, there is always a physical ultraviolet cut- II. FROM CANONICAL QUANTIZATION ON
off at which the effective low-energy quantum mechanical THE ENTIRE REAL AXIS TO AFFINE
description in the continuum breaks down, such that our QUANTIZATION ON THE HALF-LINE
momentum concept is indeed applicable. Whether the
Planck length leads to a physical ultraviolet cut-off in
extra-dimensional spaces with boundaries is, of course, a In this section we address the non-self-adjointness of
matter of speculation. the standard momentum operator on the half-line and
its consequences for canonical and affine quantization.
The paper is written also with some pedagogical in-
tentions in mind. Therefore we do not assume that the
reader is familiar with the concept of operator domains,
which is crucial for the distinction between Hermiticity
and self-adjointness. Unfortunately, in the education of A. Non-self-adjointness of −i∂x on R≥0
the typical theoretical physicist these issues often do not
play a prominent role. The experts will hopefully not be Let us consider the standard momentum operator p̂ =
offended that we elaborate on some issues that are well- −i∂x on the half-line R≥0 . Using partial integration one
known to them. We also like to point out that the notion obtains
of canonical quantization is not as uniquely defined as one
might think. Definitely, it describes quantization in an
equal-time Hamiltonian formulation, rather than, for ex- hp̂† χ|Ψi = hχ|p̂Ψi = hp̂χ|Ψi − iχ(0)∗ Ψ(0) . (1)
ample, on a light-cone or some other hyper-surface. Here
we use a more narrow definition of canonical quantiza- Hermiticity requires that χ(0)∗ Ψ(0) = 0. This re-
tion, which is based upon the canonical commutation re- quirement can be satisfied if one restricts the domain
lations between coordinates and conjugate momenta (or D(p̂) to those wave functions whose derivative is square-
closely related variants thereof). This definition distin- integrable and that obey Ψ(0) = 0. However, then χ(0)
guishes canonical quantization from affine quantization, remains unrestricted and can still assume arbitrary val-
which also operates in an equal-time Hamiltonian frame- ues. Consequently, the domain of p̂† , which acts on χ,
work, but replaces the momentum by the dilation opera- remains unrestricted, and D(p̂† ) ⊃ D(p̂). When Ψ(0) =
tor. The main new result of our work is the construction 0, p̂ and p̂† act in the same way and hence p̂ = −i∂x
of an appropriate self-adjoint momentum operator, which is indeed Hermitean. However, since D(p̂† ) 6= D(p̂), it
forms the basis for successfully applying canonical quan- is not self-adjoint. In fact, it is impossible to extend p̂
tization to the half-line and to an interval, for which it to a self-adjoint operator on the half-line. Consequently,
was thought to be inapplicable. Still, applying canon- p̂ = −i∂x does not describe the physical momentum of a
ical quantization (defined in this way) to the half-line quantum mechanical particle that moves along the pos-
or an interval is less straightforward than for the entire itive real axis. As a result, it has been concluded that,
real axis, because some subtleties related to operator do- in this case, momentum is no longer an observable phys-
mains are crucial. Although, this is well-known to the ical quantity [8]. We will reach a different conclusion,
experts, we will discuss explicitly how canonical quanti- namely that not p̂ = −i∂x (which is not self-adjoint) but
zation should be applied in such cases. another operator, p̂R = −iσ1 ∂x , which is self-adjoint in
The rest of the paper is organized as follows. In Sec- the Hilbert space L2 (R2≥0 ) of the doubly-covered positive
tion II, we address the non-self-adjointness of the stan- real axis, describes the physical momentum of a particle
dard momentum operator p̂ = −i∂x on the half-line R≥0 , on the half-line. In fact, the appropriate momentum op-
we construct the self-adjoint extensions of the Hamilto- erator p̂R + ip̂I has a Hermitean component p̂R as well
nian, and we address canonical quantization and stan- as an anti-Hermitean component ip̂I , with both p̂R and
dard momentum measurements, as well as affine quan- p̂I being self-adjoint.
4
B. Self-adjoint extension of Ĥ on R≥0 In addition, for γ < 0, there is a bound state of nega-
tive energy
When restricted to the half-line, the self-adjointness
of Ĥ = − 2m1
∂x2 + V (x), with V (x) being non-singular,
p γ2
ψb (x) = −2γ exp(γx) , Eb = − . (9)
requires the following adaptations. First of all, by per- 2m
forming two partial integrations one obtains
Interestingly, a perfectly reflecting impenetrable barrier
† can still support bound states, and the Hamiltonian Ĥ =
hĤ χ|Ψi = hχ|ĤΨi =
1
1 − 2m ∂x2 (endowed with a negative self-adjoint extension
hĤχ|Ψi − [∂x χ(0)∗ Ψ(0) − χ(0)∗ ∂x Ψ(0)] . (2) parameter γ < 0) indeed has an eigenstate of negative
2m
energy. It is easy to convince oneself that the bound state
Hermiticity hence requires the term in square-brackets is orthogonal to the scattering states, i.e. hψE |ψb i = 0.
to vanish. The most general boundary condition that is
consistent with the linearity of quantum mechanics is the
Robin boundary condition C. Canonical quantization on R
γΨ(0) − ∂x Ψ(0) = 0 . (3)
Canonical quantization is based upon the canonical
Dirichlet boundary conditions, Ψ(0) = 0, result from γ → commutation relation [x̂, p̂] = i. This relation applies
∞, while Neumann boundary conditions, ∂x Ψ(0) = 0, to unrestricted linear motion, because in the Hilbert
correspond to γ = 0. Since (for finite γ) Ψ(0) itself can space L2 (R) the momentum operator p̂ = −i∂x is indeed
still take arbitrary values, inserting eq.(3) in the square- self-adjoint. The operator p̂ then generates infinitesimal
bracket in eq.(2), the Hermiticity condition turns into translations in coordinate space. The unitary operator
that translates a wave function by a distance −a acts as
[∂x χ(0)∗ − γχ(0)∗ ] Ψ(0) = 0 ⇒ γ ∗ χ(0) − ∂x χ(0) = 0 .
(4)
Ua Ψ(x) = exp(ip̂a)Ψ(x) = exp(a∂x )Ψ(x) = Ψ(x + a) .
This relation characterizes the domain D(Ĥ † ) of Ĥ † (10)
(which acts on χ). The Hamiltonian is self-adjoint only By a Fourier transformation, we obtain the momentum
if the two domains coincide, D(Ĥ † ) = D(Ĥ), i.e. if space wave function
γ ∗ = γ ∈ R. In this way, we obtain a 1-parameter family Z ∞
of self-adjoint extensions of Ĥ.
Ψ(p)
e = hp|Ψi = dxhp|xihx|Ψi
Self-adjointness is not just a mathematical require- −∞
ment, it also has most important physical consequences. Z ∞
In particular, for γ ∈ R the boundary condition of eq.(3) = dx exp(−ipx)Ψ(x) . (11)
ensures that the probability current density −∞
1 The position operator x̂ = i∂p generates infinitesimal

j(x) = [Ψ(x)∗ ∂x Ψ(x) − ∂x Ψ(x)∗ Ψ(x)] , (5) translations in momentum space. The unitary operator
2mi
U
eq = exp(iqx̂), which translates a momentum space wave
vanishes at the boundary, i.e. j(0) = 0, and hence does function by q, acts as
not flow into the forbidden region on the negative real
axis. This ensures unitarity, i.e. probability conservation, U
eq Ψ(p)
e = exp(−q∂p )Ψ(p)
e e − q) .
= Ψ(p (12)
for the particle moving along the half-line.
In the absence of a potential (V (x) = 0) it is easy to For momentum eigenstates hx|ki = exp(ikx) (which are
construct the energy eigenstates. First of all, there are orthonormalized to hk 0 |ki = 2πδ(k − k 0 )) we have
p2
stationary scattering states of positive energy E = 2m
hx|k + qi = exp(iqx)hx|ki = U
eq hx|ki . (13)
ip + γ
ψE (x) = exp(−ipx) + R(p) exp(ipx) , R(p) = .
ip − γ We also find
(6)
It is sufficient to limit oneself to p ≥ 0, because (together Ua U
eq Ψ(x) = Ua exp(iqx)Ψ(x) = exp(iq(x + a))Ψ(x + a),
with a bound state for γ < 0) these states alone form a
U
eq Ua Ψ(x) = exp(iqx)Ψ(x + a) . (14)
complete orthonormalized set with
hψE 0 |ψE i = 2πδ(p − p0 ) , (7) Hence, as a counterpart to the Heisenberg algebra [x̂, p̂] =
i, one obtains the Weyl group relation
02
p
where E 0 = 2m , p0 > 0. In particular, since R(−p) =
∗ −1
R(p) = R(p) , the states with opposite values of p are Ua U
eq = exp(iqa)U
eq Ua . (15)
simply given by
The action of the operators Ua and U
eq is illustrated in
exp(ipx) + R(−p) exp(−ipx) = R(−p)ψE (x) . (8) Fig.1.
5
|x − ai |xi γ → 0− becomes proportional to the step function θ(x),

x̂ is not identical with the k = 0 momentum eigenstate,
which is constant over the entire real axis. This prop-
Ua erty results from the fact that |ψb i belongs to the Hilbert
space L2 (R≥0 ) while |ki resides in an extension of L2 (R).
|ki |k + qi
p̂ Next, we consider standard momentum measurements
performed on the positive energy scattering state |ψE i
p2
U
eq with E = 2m , p ≥ 0. It is straightforward to obtain
FIG. 1. Action of the translation operators Ua = exp(ip̂a) and
U
eq = exp(iqx̂) on the position and momentum eigenstates for
1 1

the entire real axis R. hk|ψE i = −i lim+ + R(p) .
→0 k − i + p k − i − p
(19)
D. Standard momentum measurements Using the residue theorem one then confirms that
Since the standard momentum operator p̂ = −i∂x is 1

Z ∞
self-adjoint only over the entire real axis but not over the hψE 0 |ψE i = dk hψE 0 |kihk|ψE i = 2πδ(p − p0 ) .
2π −∞
half-line, applying the standard concept of a momentum (20)
measurement requires to extend the Hilbert space from In this case, one might expect that the only possible
L2 (R≥0 ) to L2 (R) [41]. This can be achieved by assigning measurement results for the standard momentum are
a finite constant potential V (x < 0) = V0 to the negative k = p and k = −p, each with probability 12 . However,
real axis and finally sending V0 → ∞ [42]. A momentum as we will see at the end of the paper, these probabili-
measurement then projects the wave function (which is ties are only 14 . Due to the two different Hilbert spaces
exponentially suppressed on the negative real axis) on
L2 (R≥0 ) and L2 (R), ψE (x), which vanishes for x < 0,
a plane wave hx|ki = exp(ikx), which is unsuppressed
is not a linear combination of hx| − pi = exp(−ipx) and
for x < 0. Such a momentum measurement catapults
R(p)hx|pi = R(p) exp(ipx) along the entire real axis.
the particle out of the energetically allowed region and
transfers an infinite amount of energy to the particle in We may conclude that it is possible to enforce the stan-
the limit V0 → ∞. Such a measurement can be real- dard concept of momentum for a particle on the half-line,
ized, for example, in an optical box trap, if the ultra-cold however, at the price of putting the particle also onto the
atoms are released from the trap immediately before the negative real axis as a result of the measurement. We
momentum measurement. will soon present an alternative concept for a self-adjoint
Let us first consider the bound state |ψb i for γ < 0. momentum operator, which actually leads to the same
Its overlap with the eigenstates |ki determines the prob- probability distribution for the measurement results, but
ability density to obtain the value k in a measurement of strictly confines the particle to the positive real axis also
the standard momentum operator after a measurement.
1 1 γ
|hk|ψb i|2 = − . (16)
2π π γ 2 + k2
As expected, the resulting momentum expectation value

vanishes. The momentum uncertainty diverges because E. Affine quantization on R≥0
Z ∞
1 ∞ γk 2
Z
1
dk k 2 |hk|ψb i|2 = − dk 2 =∞. Since on the half-line there is no translation invariance,
2π −∞ π −∞ γ + k2
the operator p̂ = −i∂x is not self-adjoint. As a result, the
(17) standard commutation relation [x̂, p̂] = i that underlies
In the limit γ → 0− the bound state becomes an unbound canonical quantization has only a formal status and is
zero-energy scattering state, and the probability density no longer physically meaningful. Consequently, in this
to measure the momentum k turns into case canonical quantization based upon p̂ fails. For this
1 γ reason, on the half-line affine quantization has played
− lim = δ(k) . (18) an important role [36–40]. The momentum operator p̂ is
γ→0− π γ 2 + k2
then replaced by the generator dˆ of infinitesimal dilations.
This seems to suggest that, in this limit, the value k = 0 The unitary operator that dilates a wave function by a
is measured with certainty, in contradiction to the diver- scale factor 1/s ∈ R>0 acts as
gent uncertainty that we found for γ < 0. This shows
that the limit γ → 0− is not approached uniformly. In √
any case, the state ψb (x) = Aθ(x), which in the limit Us Ψ(x) = exp(idˆlog s)Ψ(x) = sΨ(sx) . (21)
6
Let us consider an infinitesimal dilation s = 1 + with III. FROM A NEW CONCEPT FOR THE
small . We then obtain MOMENTUM OPERATOR TO CANONICAL
√ QUANTIZATION ON THE HALF-LINE
Us Ψ(x) = 1 + Ψ(x + x)

≈ 1+ (Ψ(x) + x∂x Ψ(x)) In this section, we introduce a new concept for the
2
momentum operator in a space with sharp boundaries,
ˆ
≈ (1 + id)Ψ(x) ⇒
which allows us to apply canonical quantization to the
ˆ 1 half-line.
dΨ(x) = −i + x∂x Ψ(x) ⇒
2

ˆ 1 i
d = −i + x∂x = − (∂x x + x∂x ) . (22) A. A self-adjoint momentum operator on the
2 2 half-line
The commutation relation that replaces [x̂, p̂] = i in affine
quantization is Recently we have developed a new concept for the
ˆ = [x̂, −ix∂x ] = ix̂ . momentum of a quantum mechanical particle in a 1-d
[x̂, d] (23)
box [− L2 , L2 ] [9], which readily extends to the half-line as
By performing a partial integration we obtain well as to higher-dimensional spaces with sharp bound-
aries. For the convenience of the reader, we summa-
hdˆ† χ|Ψi = hχ|dΨi
ˆ
rize the most important aspects of this construction in
Z ∞
∗ 1 Appendix A. The construction results from the contin-
= dx χ(x) −i + x∂x Ψ(x)
0 2 uum limit of a system that is regularized on a spatial
Z ∞
1
∗ lattice. On a lattice, the derivative that enters the mo-
= dx −i + x∂x χ(x) Ψ(x) mentum operator is replaced by a nearest-neighbor finite
0 2
difference. One must distinguish forward and backward
ˆ†
= hd χ|Ψi . (24) derivatives, neither of them being Hermitean. Only the
symmetrized forward-backward derivative, which corre-
As long as xχ(x)∗ Ψ(x) vanishes at the boundary x = 0, dˆ
sponds to a next-to-nearest neighbor finite difference that
is Hermitean. Since square-integrable
√ wave functions are
extends over two lattice spacings, results in a Hermitean
less singular than 1/ x near the origin, this condition is
momentum operator. The lattice is naturally divided
satisfied without further domain restrictions. As a result,
ˆ = D(dˆ† ) and dˆ is indeed self-adjoint. into two sublattices, one with even and one with odd lat-
D(d) tice sites. The symmetrized forward-backward derivative
The eigenfunctions of dˆ obey associated with an even site then results from the values

ˆ 1 of the wave function at the two neighboring odd sites. In
d|κi = −i + x∂x |κi = κ|κi ⇒ the continuum limit, the sublattice structure naturally
2
leads to a two-component wave function, on which the
1 x iκ−1/2
hx|κi = √ . (25) momentum operator acts as a 2 × 2 matrix
l l
Here l is an arbitrarily chosen fixed length scale. Similar 0 ∂x Ψe (x)
p̂R = −i = −iσ1 ∂x , Ψ(x) = .
to the momentum eigenstates |ki over R, the eigenstates ∂x 0 Ψo (x)
|κi of dˆ are not normalizable in the usual sense. The ana- (29)
log of Fourier transformation for canonical quantization As a result, the problem is elevated to the Hilbert space
is a Mellin-type transformation [36] for affine quantiza- L2 (R2≥0 ) of square-integrable functions on the double-
tion cover of the half-line. This is the crucial insight that leads
Z ∞ to the construction of a self-adjoint momentum operator.
Ψ(κ)
e = hκ|Ψi = dx hκ|xihx|Ψi The full momentum operator p̂R + ip̂I has both a Her-
0 mitean component p̂R (which can be extended to a self-
Z ∞
1 x −iκ−1/2 adjoint operator) and an anti-Hermitean component ip̂I
= dx √ Ψ(x) . (26)
0 l l with
Let us also introduce the unitary operator U eδ =
1 δ(x − ) 0
exp(iδ x̂) which leads to p̂I = lim . (30)
2 →0+ 0 0
√
Us U
eδ Ψ(x) = Us exp(iδx)Ψ(x) = s exp(iδsx)Ψ(sx) ,
√ The operator p̂I is self-adjoint and diagonal in the posi-
U
eδs Us Ψ(x) = exp(iδsx) sΨ(sx) . (27) tion basis.
Hence, in analogy to the Weyl group for canonical quan- Let us first investigate the Hermiticity of p̂R . By par-
tization, for affine quantization one obtains the affine tial integration we obtain
group relation hp̂†R χ|Ψi = hχ|p̂R Ψi =
Us U
eδ = U
eδs Us . (28) hp̂R χ|Ψi + i[χe (0)∗ Ψo (0) + χo (0)∗ Ψe (0)] . (31)
7
Next, we impose the boundary condition What is the most general boundary condition at x = 0
for the extended Hamiltonian Ĥ(µ)? The linearity of
Ψo (0) = λΨe (0) , (32)
quantum mechanics restricts us to write
which constrains the domain D(p̂R ). Inserting this rela-
Ψo (0) iη a −b Ψe (0)
tion in eq.(31), the Hermiticity condition becomes =e . (38)
∂x Ψo (0) −c d ∂x Ψe (0)
[χe (0)∗ λ + χo (0)∗ ]Ψe (0) = 0 . (33)
Self-adjointness of Ĥ(µ) again demands that the prob-
Since Ψe (0) can still take arbitrary values, one obtains ability current density vanishes at the origin, j(0) = 0.
χo (0) = −λ∗ χe (0) . (34) For 2-component wave functions the current density takes
the form
The operator p̂R is self-adjoint if D(p̂†R ) = D(p̂R ), which 1
is true when λ = −λ∗ , such that λ ∈ iR. As a j(x) = [Ψ(x)∗ ∂x Ψ(x) − ∂x Ψ(x)∗ Ψ(x)]
2mi
result, we obtain a 1-parameter family of self-adjoint 1
extensions, characterized by the purely imaginary pa- = [Ψe (x)∗ ∂x Ψe (x) − ∂x Ψe (x)∗ Ψe (x)
2mi
rameter λ. Since p̂R is self-adjoint, it has a complete + Ψo (x)∗ ∂x Ψo (x) − ∂x Ψo (x)∗ Ψo (x)] . (39)
set of orthonormal eigenstates with corresponding real
eigenvalues. The momentum eigenstates, which obey Using j(0) = 0, it is straightforward to derive the con-
p̂R φk (x) = kφk (x) with k ∈ R, are given by ditions a, b, c, d ∈ R and ad − bc = −1. Together
with η, these are four independent parameters, which
1 exp(ikx) + σ exp(−ikx) define a family of self-adjoint extensions. In order to
φk (x) = √ ,
2 exp(ikx) − σ exp(−ikx) correctly embed the original Hamiltonian Ĥ, the bound-
1−λ ary condition must support the finite-energy states with
σ= ∈ U (1) . (35) Ψ+ +
1+λ o (x) = Ψe (x), which requires
Like the momentum eigenstates |ki on the entire real eiη = 1 , a=1, b=0, d = −1 . (40)
axis, the states |φk i on the half-line are orthonormalized
to δ-functions, i.e. hφk0 |φk i = 2πδ(k − k 0 ). Using these specific parameters, eq.(38) reduces to
c
− Ψ+ (0) − ∂x Ψ+ (0) = 0 , Ψ− (0) = 0 . (41)
2
B. Embedding of Ĥ in L2 (R2≥0 ) We now identify Ψ+ (x) with the wave functions in the
original Hilbert space L2 (R≥0 ). Putting γ = −c/2,
The search for a self-adjoint momentum operator that eq.(41) reduces to the Robin boundary condition of
is strictly limited to the half-line has naturally put us into eq.(3), while the wave functions Ψ− (x) obey Dirich-
the Hilbert space L2 (R2≥0 ) of the doubly covered positive let boundary conditions. By construction, in the limit
real axis. The double cover reflects the importance of µ → ∞ the Hamiltonian Ĥ(µ) has the same finite-energy
even and odd lattice points in the underlying ultraviolet spectrum as the original Hamiltonian Ĥ. The corre-
regularization at the level of the continuum description sponding eigenstates of Ĥ(µ) are just identical copies of
that emerges in the limit of vanishing lattice spacing. In the original eigenstates of Ĥ (renormalized by √12 ) in the
order to apply the new concept for the momentum to
upper and lower component of the 2-component wave
the particle on the half-line, we must embed the original
1 function.
Hamiltonian Ĥ = − 2m ∂x2 + V (x) with the self-adjoint
extension parameter γ, which acts in L2 (R≥0 ), into the
doubled Hilbert space L2 (R2≥0 ). This is achieved by con- C. Momentum measurements on R≥0
structing the Hamiltonian
1

− 2m ∂x2 + V (x) 0 We are now ready to apply the new concept of momen-
Ĥ(µ) = 1 + µP̂− .
0 − 2m ∂x2 + V (x) tum to the particle on the half-line. An original wave
(36) function ψ(x) ∈ L2 (R≥0 ) is trivially embedded in the
Here P̂− projects on states Ψ− (x) with Ψ− (x) = doubled Hilbert space L2 (R≥0 )2 as
o
−Ψ− e (x). In the underlying lattice theory, these states +
Ψe (x)

1

ψ(x)

+
have energies at the lattice cut-off. In order to decou- Ψ (x) = =√ . (42)
Ψ+o (x) 2 ψ(x)
ple them from the continuum theory, we take the limit
µ → ∞. The complementary operator P̂+ projects on The probability to measure a momentum value k is de-
the remaining states Ψ+ (x) with Ψ+ +
o (x) = Ψe (x), which
termined by the amplitude
have finite energy, i.e. †
1 ∞
Z
+ exp(ikx) + σ exp(−ikx) ψ(x)
1

1 ±1
hφk |Ψ i = dx
exp(ikx) − σ exp(−ikx) ψ(x)
P̂± = , P̂±2 = P̂± , P̂+ + P̂− = 1 , 2 0
2 ±1 1 Z ∞
= dx exp(−ikx)ψ(x)θ(x) = hk|ψi . (43)
Ψ(x) = Ψ+ (x) + Ψ− (x) , Ψ± (x) = P̂± Ψ(x) . (37) −∞
8
Remarkably, hk|ψi is just the amplitude that determines obey the canonical commutation relation [x̂R , p̂R ] = i.
the probability to obtain the value k in a measurement According to the Stone-von-Neumann theorem [43–46],
of the standard momentum operator p̂ = −i∂x that is this implies that x̂R and p̂R are unitarily equivalent to the
self-adjoint only over the entire real axis. Here, using the standard coordinate x̂ and momentum p̂. While this is
step function θ(x), the wave function ψ(x) has been triv- true mathematically, if not properly interpreted, it might
ially extended to the negative real axis. As a result, both be physically misleading. In particular, while the eigen-
the standard and the new concept of momentum yield states of x̂ describe positions on the entire real axis, the
the same probability distributions for the measurement eigenstates of x̂R describe a double-cover of the half-line.
results. However, the two concepts project onto differ- Hence, the situations are, in fact, physically distinct, but,
ent states after the measurements. In particular, while a as we will see explicitly below, indeed mathematically re-
standard momentum measurement puts the particle also lated by a unitary transformation. This may still seem
onto the negative real axis, the new concept is strictly surprising, since p̂R possesses the self-adjoint extension
limited to the half-line, i.e. also after the measurement parameter λ ∈ iR which characterizes the domain D(p̂R ),
the particle remains on the positive real axis. Still, also while the standard momentum operator does not. How-
with the new concept a momentum measurement trans- ever, two operators p̂R , which are endowed with different
fers an infinite amount of energy to the particle. This is self-adjoint extension parameters λ and λ0 , are indeed
because the momentum eigenstate φk (x), onto which a related by a unitary transformation
measurement that results in the value k projects, not only
has the finite-energy component φ+ √1 exp(ikx)

k (x) = 2 W = exp(iωσ1 ) =
cos ω i sin ω
,
but also the component φ− √1 σ exp(−ikx), whose i sin ω cos ω
k (x) = 2
energy diverges in the limit µ → ∞. Interestingly, W (−iσ1 ∂x )W † = −iσ1 ∂x , (45)
the purely imaginary self-adjoint extension parameter λ,
which determines σ = (1 − λ)/(1 + λ) ∈ U (1), does not which leaves the differential expression for p̂R invariant.
affect the probability to obtain a certain momentum mea- The boundary condition Ψo (0) = λΨe (0) then turns into
surement result.
Ψ0e (0)

As we will see later, the standard and the new concept Ψe (0) cos ωΨe (0) + i sin ωΨo (0)
=W =
for the momentum no longer result in the same mea- Ψ0o (0) Ψo (0) i sin ωΨe (0) + cos ωΨo (0)
surement results when one considers a particle that is
(cos ω + λi sin ω)Ψe (0)
confined to an interval [9]. In that case, a standard mo- = ⇒
(i sin ω + λ cos ω)Ψe (0)
mentum measurement catapults the particle outside of
the interval and results in a continuous momentum value. i sin ω + λ cos ω 0
Ψ0o (0) = Ψ (0) = λ0 Ψ0e (0) . (46)
The new concept, on the other hand, yields discrete mo- cos ω + λi sin ω e
mentum values and leaves the particle inside the interval
after a measurement. As a result, two different operators p̂R , which are associ-
ated with the self-adjoint extension parameters λ and λ0
are related by the unitary transformation W with
D. Canonical quantization on the half-line R≥0
λ − λ0
tan ω = i ,
As we have seen in eq.(10), in canonical quantization 1 − λλ0
the unitary operator Ua = exp(ip̂a) with p̂ = −i∂x per- 1 − λ0 1−λ
σ0 = = exp(−2iω) = exp(−2iω)σ . (47)
forms a coordinate shift, Ua Ψ(x) = Ψ(x + a), while the 1+λ 0 1+λ
operator U eq = exp(iqx̂) with x̂ = i∂p performs a mo-
mentum shift, U eq Ψ(p)
e e − q) (cf. eq.(12)). What is
= Ψ(p Let us again denote the eigenstates of x̂ as |xi, such
that x̂|xi = x|xi. Here we limit ourselves to the positive
the unitary operator Veq that shifts the eigenstates φk (x)
real axis with x > 0. The eigenstates of x̂R = σ1 x are
of the self-adjoint momentum operator on the half-line,
then given by
p̂R = −iσ1 ∂x , to φk+q (x)? We construct

1 exp(i(k + q)x) + σ exp(−i(k + q)x) 1 |xi
φk+q (x) = √ x̂R |x, +i = x|x, +i , |x, +i = √ ,
2 exp(i(k + q)x) − σ exp(−i(k + q)x) 2 |xi

cos(qx) i sin(qx) 1 exp(ikx) + σ exp(−ikx) 1 |xi
= √ x̂R |x, −i = −x|x, −i , |x, −i = √ . (48)
i sin(qx) cos(qx) 2 exp(ikx) − σ exp(−ikx) 2 −|xi
= Veq φk (x) ⇒
The eigenstates |x, +i have a positive eigenvalue x > 0
cos(qx) i sin(qx) and belong to the finite-energy sector, i.e. P̂+ |x, +i =
Veq = = exp(iqx̂R ), x̂R = σ1 x. (44)
i sin(qx) cos(qx) |x, +i, while the eigenstates |x, −i have a negative eigen-
Not surprisingly, the conjugate coordinate to the momen- value −x and obey P̂− |x, −i = |x, −i, P̂+ |x, −i = 0.
tum p̂R = −iσ1 ∂x is x̂R = σ1 x. Indeed, the two operators How does the operator Va = exp(ip̂R a) act on the
9
Va |x, +i x̂R and p̂R with x̂ and p̂. The corresponding unitary
x̂R transformation U is given by
0
|a − x, −i U Va U † = Ua , U Veq U † = U
eq ,
U x̂R U † = x̂ , U |x, +i = |xi , U |x, −i = σ ∗ | − xi ,
|ki |k + qi
U p̂R U † = p̂ , U |φk i = |ki . (52)
p̂R
We like to stress again that this mathematical unitary
Veq transformation relates two physically very different situ-
ations. In particular, while the states |x, +i (with x > 0)
FIG. 2. Action of the translation operators Va = exp(ip̂R a) in the finite-energy sector are mapped to position eigen-
and Veq = exp(iqx̂R ) on the position and momentum eigenstates |xi on the positive real axis, the states |x, −i (again
states for the half-line R≥0 . with x > 0), which are mapped to the states | − xi on the
negative real axis, actually reside on the positive half-
line, but have ultraviolet-sensitive energies and are re-
eigenstates |x, ±i? We obtain moved from the spectrum of the Hamiltonian Ĥ(µ) in
Z ∞ the limit µ → ∞.
1
Va |x, ±i = dk exp(ip̂R a)|φk ihφk |x, ±i
2π −∞
Z ∞ E. The classical limit
1
= dk exp(ika)|φk ihφk |x, ±i ,
2π −∞
Until now we have used the term quantization — be
1 exp(ikx) + σ exp(−ikx) it canonical or affine — without even starting from a
hx, +|φk i = (1, 1)
2 exp(ikx) − σ exp(−ikx) classical theory. Since quantum physics is more funda-
= exp(ikx) , mental than classical physics, this way to proceed is not
1

exp(ikx) + σ exp(−ikx)
unreasonable. Until now, we have shown that our new
hx, −|φk i = (1, −1) concept gives rise to a self-adjoint momentum operator
2 exp(ikx) − σ exp(−ikx)
p̂R = −iσ1 ∂x that obeys the canonical commutation re-
= σ exp(−ikx) , (49) lation, [x̂R , p̂R ] = i, with the corresponding coordinate
x̂R = σ1 x. We now consider the classical limit in order
which implies
to demonstrate that the new concept still makes sense
Va |x, +i = |x − a, +i , for 0 ≤ x − a , when commutators are replaced by Poisson brackets.
Let us perform a unitary transformation D that diag-
Va |x, +i = σ|a − x, −i , for x − a < 0 , onalizes x̂R , p̂R , and Ĥ(µ) such that
Va |x, −i = |x + a, −i , for 0 ≤ x + a ,
Va |x, −i = σ ∗ | − x − a, +i , for x + a < 0 . 1 1 1
(50) Dσ1 D† = σ3 , D = √ ,
2 1 −1
Writing x̂R |xR i = xR |xR i, xR ∈ R, we identify |xR i =

x 0
|xR , +i for xR > 0, and |xR i = | − xR , −i for xR < x̂0R = Dx̂R D† = xσ3 = , x ∈ R≥0 ,
0 −x
0. Eq.(50) then implies that, up to a phase σ or σ ∗ ,
Va shifts the eigenstates of x̂R from |xR i to |xR − ai. ∂x 0
p̂0R = Dp̂R D† = −iσ3 ∂x = −i ,
Independent of σ, as a consequence of the Heisenberg 0 −∂x
algebra [x̂R , p̂R ] = i, one obtains the Weyl group relation Ĥ(µ)0 = DĤ(µ)D†
1

− 2m ∂x2 + V (x) 0
Va Veq = exp(iqa)Veq Va . (51) = 1 . (53)
0 − 2m ∂x2 + V (x) + µ
The action of the operators Va and Veq is illustrated in The positive eigenvalues of x̂R correspond to physical
Fig.2. The operator Va = exp(ip̂R a) transports states points x ∈ R≥0 on the half-line on which finite-energy
|x, +i in the finite-energy sector to the left, and states states are located. The negative eigenvalues −x of x̂R ,
|x, −i, whose energy is ultraviolet-sensitive, to the right. on the other hand, are not associated with the negative
When a state |x, +i is transported beyond the origin (at real axis, but host states with energies at the ultraviolet
x = 0), it turns into the state |a − x, −i. cut-off scale, which are removed in the limit µ → ∞.
For mathematical purposes, the Weyl groups are more A similar mathematical description arises when one
convenient than the corresponding Heisenberg algebras applies the standard concept of momentum. In that
because unitary operators are bounded while the cor- case, one regularizes the problem by assigning a potential
responding self-adjoint operators are not. Based upon V (x < 0) = V0 to points on the negative real axis, and
the Weyl groups of eq.(15) and eq.(51), the Stone-von- ultimately taking the limit V0 → ∞. At a formal level V0
Neumann theorem guarantees the unitary equivalence of then corresponds to µ, but the physical interpretation is
10
very different. In fact, when one applies the new concept γ < ∞, one then has Ψe (0) = Ψo (0), which is inconsis-
for the momentum, the particle is strictly limited to the tent with Ψo (0) = λΨe (0), because λ ∈ iR is purely imag-
positive real axis, even after a momentum measurement. inary. As a result, the domains of the momentum oper-
In any case, one can now replace the commutation re- ator and the kinetic energy operator are not the same,
lation [x̂0R , p̂0R ] = i by the classical Poisson bracket re- D(p̂R ) 6= D(T̂ ). Only when we choose Dirichlet bound-
lation {x, p} = 1 and replace the Hamiltonian Ĥ(µ)0 ary conditions (which correspond to γ → ∞), the domain
in the limit µ → ∞ by the classical Hamilton function of T̂ is characterized by Ψe (0) = Ψo (0) = 0, which is au-
p2 tomatically consistent with Ψo (0) = λΨe (0). Hence, in
H(x, p) = 2m + V (x). The hard wall at x = 0 arises as
a consequence of µ → ∞, thus leading to the standard this special case, D(T̂ ) ⊂ D(p̂R ). This implies that an
motion of a classical particle that gets reflected at the ori- application of p̂R on a wave function T̂ Ψ(x) ∈ D(T̂ ) is
gin. The self-adjoint extension parameters λ and γ leave still possible and p̂R T̂ Ψ(x) ∈ D(p̂R ). On the other hand,
no trace in the classical limit. In particular, unlike the an application of T̂ on p̂R Ψ(x) ∈ D(p̂R ) is possible only
quantum particle, the classical particle cannot be bound if p̂R Ψ(x) ∈ D(T̂ ). Hence, the commutator p̂R T̂ − T̂ p̂R
to the reflecting wall at x = 0.
cannot act on all wave functions in D(p̂R ) or D(T̂ ), and
hence has only a limited formal meaning. As a result,
F. Subtleties with commutators and operator although the operators p̂R and T̂ seem to commute at
domains the superficial level of differential expressions, they do
not have common eigenfunctions.
Canonical quantization (as defined in the introduction) Let us return to the operators x̂ and dˆ = −i( 21 + x∂x )
usually starts from a classical theory with the Poisson that we encountered in affine quantization and consider
bracket relation {x, p} = 1, which is then promoted to them together with the operators x̂R = σ1 x and p̂R =
the commutation relation [x̂, p̂] = i between self-adjoint −iσ1 ∂x that we used in canonical quantization. Among
operators x̂ and p̂. Until now, it was assumed that on those, only the domain D(p̂R ) is further restricted by the
the half-line canonical quantization fails, simply because condition Ψo (0) = λΨe (0) with λ ∈ iR. First of all,
in this case no self-adjoint momentum operator exists for

0 x Ψe (x) xΨo (x)
L2 (R≥0 ). As we have seen, a satisfactory self-adjoint mo- x̂R Ψ(x) = = , (55)
x 0 Ψo (x) xΨe (x)
mentum operator p̂R = −iσ1 ∂x as well as its canonically
conjugate coordinate x̂R = σ1 x (with x ≥ 0) indeed ex- which does not lead out of D(p̂R ), because xΨe (x) =
ist in the doubled Hilbert space L2 (R2≥0 ). The operator λxΨo (x) = 0 at x = 0. Trivially embedding x̂ into the
p̂R is endowed with the self-adjoint extension parame- doubled Hilbert space, we also obtain
ter λ, and is thus not uniquely defined. However, as we
x̂ 0 Ψe (x) xΨe (x)
have seen in eq.(46), in accordance with the Stone-von- = , (56)
0 x̂ Ψo (x) xΨo (x)
Neumann theorem [43–46], two operators p̂R , which are
associated with two different self-adjoint extension pa- which does not lead out of D(p̂R ) either. Finally, by also
rameters λ and λ0 , are, in fact, unitarily equivalent. embedding dˆ into the doubled Hilbert space, one obtains
Following the canonical quantization procedure, one
p2 dˆ 0 −i( 21 + x∂x )Ψe (x)

replaces the classical Hamilton function H(x, p) = 2m + Ψe (x)
= . (57)
V (x) with a Hamilton operator Ĥ by replacing x with 0 dˆ Ψo (x) −i( 21 + x∂x )Ψo (x)
x̂ and p with p̂. When one replaces the classical kinetic Putting x = 0, we then find
p2
energy T = 2m by the operator
−i( 21 + x∂x )Ψo (0) = − 2i Ψo (0) = − 2i λΨe (0)
p̂2R 1 1 2 = −λi( 12 + x∂x )Ψe (0) . (58)
T̂ = = (−iσ1 ∂x )2 = − ∂ 1, (54)
2m 2m 2m x
We thus conclude that dˆ does not lead out of D(p̂R ) ei-
at least at the formal level of differential expressions one ther. Since x̂, x̂R , and dˆ are not subject to further do-
obtains the correct kinetic energy operator that enters main restrictions of their own, limiting ourselves to wave
the Hamiltonian Ĥ(µ) of eq.(36). functions Ψ(x) ∈ D(p̂R ), the following commutators are
At a superficial level one might conclude that T̂ and indeed not just formal expressions, but mathematically
p̂R commute. However, T̂ is self-adjoint in L2 (R2≥0 ) and physically meaningful
only if it is endowed with its own self-adjoint exten-
sion parameters which characterize the domain D(T̂ ). ˆ = ix̂ , [x̂, p̂R ] = iσ1 ,
[x̂, x̂R ] = 0 , [x̂, d]
The most general self-adjoint extension of T̂ is char- ˆ = [σ1 x, −ix∂x ] = iσ1 x = ix̂R , [x̂R , p̂R ] = i ,
[x̂R , d]
acterized by eq.(38), which contains four independent ˆ = [−iσ1 ∂x , −ix∂x ] = −σ1 ∂x = −ip̂R .
[p̂R , d] (59)
parameters: η, a, b, c, d ∈ R subject to the constraint
ad − bc = −1. However, for the reasons explained above, This implies that affine and canonical quantization are
here we are only interested in the special case η = 0, fully consistent with each other and can be implemented
a = 1, b = 0, c = −2γ, d = −1. For general finite simultaneously.
11
G. Summary of the canonical quantization The remaining steps for addressing the quantum dy-
procedure namics are the usual ones, but they are strongly affected
by the selected operator domains:
To summarize, the resulting canonical quantization
• Employ the Hamiltonian to determine the wave
procedure consists of the following steps:
function that solves the Schrödinger equation.
• Select a classical system characterized by coordi- • Solve the eigenvalue problem for the operators that
nates x and canonically conjugate momenta p as- describe observables to be measured.
sociated with a classical Hamilton function H(x, p)
as well as with some observables A(x, p). • Project the wave function on the eigenfunction of
an observable in order to predict the probability to
• Identify an appropriate Hilbert space in which to measure the corresponding eigenvalue.
realize the time-evolution and the measurements
for a corresponding quantum system. It should be stressed that there is absolutely nothing
wrong with affine quantization applied to the half-line.
• Replace classical Poisson bracket relations by for- In view of our construction, it is just no longer the only
mal commutation relations. quantization procedure that is available in this case.
• Realize the commutation relations by differential

expressions. H. Is canonical quantization a fine quantization for
the half-line?
• Extend the differential expressions to self-adjoint
operators. Is there anything wrong with the above canonical
• Make a specific choice of self-adjoint extension pa- quantization procedure applied to the half-line? We will
rameters and thus of the corresponding operator now argue that this admittedly somewhat subtle proce-
domains. dure is completely fine, and, in fact, very natural in a
continuous space with sharp boundaries. First of all,
• Return to the formal commutation relations, inves- in order to motivate the construction of p̂R in the first
tigate whether they are affected by domain incom- place, in [9] we have introduced an ultraviolet regular-
patibilities, and take this into account properly. ization by replacing the continuum with a spatial lattice
with N points. The resulting Hilbert space is then N -
• If the canonical commutation relation between po- dimensional. In a finite-dimensional Hilbert space, there
sition and momentum itself is compromised by do- is no difference between Hermiticity and self-adjointness,
main incompatibilities, try to replace the position and one need not worry about the domains of operators.
or the momentum operator by more appropriate The lattice variant of the momentum operator p̂R results
operators that are unaffected by such subtleties. from a symmetrized forward-backward lattice derivative
that extends over two lattice spacings. The lattice variant
As we have seen, the identification of an appropriate of the kinetic energy operator T̂ is described by a stan-
Hilbert space may be non-trivial and may depend on dard finite-difference lattice version of a second deriva-
the set of observables to be measured. In particular, tive. The low-energy physics of the lattice system does
for the particle on the half-line the time-evolution driven not require an extension of the Hilbert space and does
by the Hamiltonian Ĥ can be realized in the Hilbert not depend on any operator domain issues. If we inter-
space L2 (R≥0 ). However, in order to consider momen- pret the lattice spacing as mimicking a shortest relevant
tum measurements, the Hilbert space must be doubled to physical length scale, such as a crystal lattice spacing in a
L2 (R2≥0 ) and the Hamiltonian must be extended accord- quantum dot, or the Planck length in an interval of extra-
ingly to Ĥ(µ). In the absence of subtle Hilbert space or dimensional space, every Hermitean operator would au-
domain incompatibility issues, all relevant commutators tomatically be self-adjoint and could act in the entire
correctly reflect the relations between the various oper- Hilbert space. In particular, the mathematical subtleties
ators, and the procedure outlined above simplifies con- related to operator domains would have no effect on the
siderably. When such subtleties do arise, on the other physics.
hand, naively applying the commutation relations that However, also from a physics point of view, it is desir-
result from classical Poisson brackets may lead to wrong able to derive a low-energy effective description of the un-
results. This would happen, for example, if one would derlying lattice physics, by taking the continuum limit of
not realize that the self-adjoint operators p̂R and T̂ do vanishing lattice spacing. This would have been straight-
not commute, although the corresponding differential ex- forward if we would care only about the spectrum of the
pressions seem to suggest this. In such a case, it is neces- Hamiltonian. In fact, one could then stay within the
sary to properly address the subtleties, which is possible Hilbert space L2 (R≥0 ). Since we also care about the mo-
using the well-developed mathematical theory that was mentum operator, the situation is more subtle. In partic-
initiated by von Neumann [27]. ular, the crucial concept of even and odd lattice points
12
naturally leads to the doubled Hilbert space L2 (R≥0 ). obeys the condition
As we have seen, the embedding of the original Hamil-
tonian Ĥ into this framework leads to the Hamiltonian (1 − λ)(1 + λL ) ∗
exp(2ikL) = = σσL = exp(iθ) . (61)
Ĥ(µ) with the specific self-adjoint extension parameters (1 + λ)(1 − λL )
of eq.(40). The limit µ → ∞ removes those states from
the finite-energy spectrum whose lattice variants have en- For λ = λL this leads to k = πn
L , n ∈ Z, while in general
ergies at the cut-off scale. This inevitably leads to the
different operator domains, D(p̂R ) 6= D(Ĥ(µ)). π θ
k= n+ , n∈Z. (62)
While the resulting canonical quantization procedure L 2π
may seem unnecessarily complicated, it is the price a
Adjacent momentum eigenvalues are hence separated by
physicist has to pay for using a low-energy continuum de- π
scription in an infinite-dimensional Hilbert space. This L . The self-adjoint extension parameter θ is invariant
against the unitary transformation W of eq.(45) because
description is, in fact, mathematically quite elegant, but
admittedly not completely straightforward. A straight- ∗
exp(iθ0 ) = σ 0 σL
0 ∗
= exp(−2iω)σ exp(2iω)σL
forward alternative description could stay on the lattice
= σσL ∗ = exp(iθ) . (63)
without taking the continuum limit. However, such a for-
mulation is not very transparent, it leads to more com- Here we have used eqs.(61) and (47).
plicated variants of the Schrödinger equation, and it does What is the unitary operator that translates the mo-
not correspond to the standard quantum mechanical con- mentum eigenfunctions by L π
? Using eq.(44), which is
tinuum description. We suggest that it is worth familiar- still applicable in the interval, we obtain
izing oneself with the differences between Hermiticity and
self-adjointness, and thus with the issues related to oper- φk+π/L (x) = Veπ/L φk (x) ,
ator domains, even if this may require an expansion of the π
cos πx i sin πx

sometimes prevailing practices in quantum mechanics. Veπ/L = exp i x̂R = L L . (64)
πx πx
L i sin L cos L
The fact that Veπ/L acts as a shift operator for the mo-
IV. CANONICAL QUANTIZATION IN AN mentum also reflects itself in the commutation relations
INTERVAL AND ON A CIRCLE π † π †
[p̂R , Veπ/L ] = Veπ/L , [p̂R , Veπ/L ] = − Veπ/L . (65)
L L
In this section, we apply the new concept for the mo-
mentum to an interval, which allows us to extend canon- Since for x = 0 and x = L the operator Veπ/L re-
ical quantization to that case as well. We then relate the duces to 1, for a wave function Ψ(x) ∈ D(p̂R ) (which
results to the well-understood situation on a circle. obeys Ψo (0) = λΨe (0) and Ψo (L) = λL Ψe (L)) one ob-
tains (Veπ/L Ψ)o (0) = λ(Veπ/L Ψ)e (0) and (Veπ/L Ψ)o (L) =
λL (Veπ/L Ψ)e (L). As a result, Veπ/L Ψ(x) ∈ D(p̂R ), which
A. Canonical quantization in the interval [0, L]
means that Veπ/L does not lead out of the domain D(p̂R ).
This implies that the commutation relations of eq.(65)
are not just formal relations, but are fully consistent with
In [9] we have introduced the new concept for the mo- the domains of the corresponding operators.
mentum operator in an interval [− L2 , L2 ]. In that work we It should be noted that the canonical commutation
have only briefly touched upon the corresponding alge- relation [x̂R , p̂R ] = i, which is completely appropriate
bra that relates the finite-volume momentum to its cor- on the half-line, is no longer a meaningful expression
responding conjugate coordinates. Here we consider this in the interval. This is because, in contrast to Veπ/L ,
problem in more detail. We shift the interval to [0, L] the operator x̂R leads out of the domain D(p̂R ). This
in order to be able to recover the half-line in the limit follows from (x̂R Ψ)e (L) = LΨo (L) = LλL Ψe (L) =
L → ∞. λL (x̂R Ψ)o (L), which is inconsistent with the domain con-
In an interval, the domain D(p̂R ) of the finite-volume dition (x̂R Ψ)o (L) = λL (x̂R Ψ)o (L). Hence, in the inter-
momentum operator p̂R is characterized by two self- val [0, L] the commutation relations of eq.(65) replace
adjoint extension parameters, λ and λL , associated with the canonical commutation relation [x̂R , p̂R ] = i. This
the two ends of the interval. In the interval [0, L] the is another example where the last step in the canonical
boundary conditions are quantization procedure is crucial: “If the canonical com-
mutation relation between position and momentum itself
Ψo (0) = λΨe (0) , Ψo (L) = λL Ψe (L) , λ, λL ∈ iR≥0 . is compromised by domain incompatibilities, try to re-
(60) place the position or the momentum operator by more
The corresponding momentum eigenfunctions φk (x) are appropriate operators that are unaffected by such sub-
still given by eq.(35). However, in contrast to the half- tleties.” Starting from the original position and momen-
line, in the interval the momentum k is quantized and tum operators x̂ and p̂ = −i∂x , which no longer obey the
13
Va |x, +i
canonical commutation relation on the half-line, we were
led to the operators x̂R = σ1 x and p̂R = −iσ1 ∂x , which
indeed satisfy [x̂R , p̂R ] = i. Moving on to an interval, we 0 L x̂R
now realize that this relation again seizes to hold, which |a − x, −i
motivated us to replace x̂R by the more appropriate op-
π π π
erator Veπ/L = exp(i L x̂R ) that obeys the commutation |L ni | L (n + 1)i
relations eq.(65). p̂R
How does the operator Va = exp(ip̂R a) act on the x̂R
eigenstates |x, ±i (with x ∈ [0, L]) of eq.(48)? In close Veπ/L
analogy to eq.(50), for a ∈ [−L, L) one obtains
FIG. 3. Action of the translation operators Va = exp(ip̂R a)
Va |x, +i = σL |a − x + 2L, −i , for L ≤ x − a < 2L , π
and Veπ/L = exp(i L x̂R ) in the continuous position and dis-
Va |x, +i = |x − a, +i , for 0 ≤ x − a < L , crete momentum eigenstates for the interval [0, L].
Va |x, +i = σ|a − x, −i , for − L ≤ x − a < 0 ,
∗
Va |x, −i = σL | − x − a + 2L, +i , for L ≤ x + a < 2L , B. Comparison with motion on S 1
Va |x, −i = |x + a, −i , for 0 ≤ x + a < L ,
Va |x, −i = σ ∗ | − x − a, +i , for − L ≤ x + a < 0 . (66) Let us compare the situation in an interval with the
well-understood motion of a quantum mechanical par-
Independent of σ or σL , as a consequence of eq.(65) one ticle on a circle S 1 , parametrized by the angle ϕ ∈
then obtains the Weyl group relation [−π, π]. In that case, the operator for the linear mo-
π mentum is replaced by the angular momentum operator
Va Veπ/L = exp i a Veπ/L Va . (67) L̂ = −i∂ϕ . First, let us investigate the Hermiticity and
L self-adjointness of L̂, which acts on wave functions Ψ(ϕ).
Applying partial integration one obtains
The action of the operators Va and Veπ/L is illustrated in
Fig.3. Since in the interval momentum is quantized, in hL̂† χ|Ψi = hχ|L̂Ψi =
the corresponding momentum space there are only dis- hL̂χ|Ψi − i[χ(π)∗ Ψ(π)−χ(−π)∗ Ψ(−π)]. (68)
π
crete translations by multiples of L , which are generated
Hermiticity of L̂ hence requires that the expression in
by Veπ/L . Despite the fact that there is no translation square brackets vanishes. Since, in contrast to an open
symmetry in the interval [0, L] (which is the reason why interval, ϕ = π and ϕ = −π parametrize the same point
the original momentum operator p̂ = −i∂x is not self- on the closed circle, the boundary condition
adjoint), the momentum operator p̂R = −iσ1 ∂x indeed
generates infinitesimal translations in the doubly covered Ψ(π) = ρΨ(−π) , (69)
interval [0, L]2 . As illustrated in Fig.3, Va = exp(ip̂R a)
which restricts the domain D(L̂), is local and thus phys-
acts like a periodic “conveyor belt”, which transports
ically admissible. Inserting this relation in the square
states |x, +i (in the finite-energy sector) to the left, and
bracket in eq.(68), and using the fact that Ψ(−π) can
states |x, −i (which lie in the sector of states whose en-
still take arbitrary values, one obtains the Hermiticity
ergy is ultraviolet-sensitive) to the right. When a state
condition
|x, +i is transported beyond the origin (at x = 0), it
turns into the state |a − x, −i. When such a state is 1
[χ(π)∗ ρ − χ(−π)∗ ]Ψ(−π) = 0 ⇒ χ(π) = ∗ χ(−π) ,
transported further beyond the other end of the interval ρ
(at x = L), it returns to the finite-energy sector. (70)
It is a matter of definition whether one wants to classify which defines the domain D(L̂† ). The operator L̂ is self-
the situation in the interval as belonging to “canonical adjoint only if D(L̂) = D(L̂† ), which is the case when
quantization”. We prefer to do so, despite the fact that 1
x̂R and p̂R no longer satisfy a meaningful canonical com- ρ= ⇒ ρ = exp(iθ) . (71)
ρ∗
mutation relation. The reason for this is simply that in a
finite volume momentum is quantized. As a consequence, It is well-known that the angular momentum operator
x̂R can no longer generate infinitesimal translations in L̂ has a 1-parameter family of self-adjoint extensions
momentum space. The appropriate discrete momentum parametrized by the angle θ. The angular momentum
translations by L π
are generated by Veπ/L = exp(i L π
x̂R ). eigenstates |ni then obey
As a result, the commutation relations of eq.(65) replace

θ
the usual canonical commutation relation. In any case, L̂|ni = n + |ni , n ∈ Z ,
2π
the corresponding Weyl group of eq.(67) has the same
form as in the other cases of canonical quantization, and 1 θ
hϕ|ni = √ exp i n + ϕ ,
thus using the term “canonical quantization” also for the 2π 2π
interval is indeed justified. hπ|ni = exp(iθ)h−π|ni . (72)
14
eigenstates |ϕi on the circle ϕ ∈ [−π, π]

|ϕi
U Va U † = Uα=πa/L , U Veπ/L U † = Ue ,
L
U x̂R U † = ϕ̂ ,
ϕ̂ π
U |x, +i = |ϕ = πx
L i , U |x, −i = σ ∗ |ϕ = − πx
L i ,
π
|ϕ − αi
U p̂R U † = L̂ , U |φk i = |n = kL
π i . (76)
L
Uα
Let us check this for consistency. For example, using
eq.(66) for L ≤ x−a < 2L, which implies π ≤ ϕ−α < 2π,
|ni |n + 1i we obtain
L̂
Va |x, +i = σL |a − x + 2L, −i ⇒
U U Va U † U |x, +i = σL U |a − x + 2L, −i ⇒
e
FIG. 4. Action of the translation operators Uα = exp(iL̂α) ∗
Uα=πa/L |ϕ = πx π
L i = σL σ |ϕ = − L (a − x + 2L)i ⇒
and U
e = exp(iϕ̂) on the continuous angle and discrete angular
π
momentum eigenstates for the circle S 1 = [−π, π]. exp(−iθ)|ϕ = L (x − a) − 2πi
= σL σ ∗ |ϕ = L π
(x − a) − 2πi . (77)
It is also well-known that ϕ̂ and L̂ do not obey the stan- This is indeed consistent, because based on eq.(61),
∗
dard canonical commutation relation. This is because an exp(iθ) = σσL .
operation with the non-periodic operator ϕ̂ leads out of Eq.(76) is very similar to eq.(52) which describes the
the domain of D(L̂). Instead, the periodic unitary oper- mathematical unitary equivalence of the standard oper-
ators x̂ and p̂ acting on wave functions over the entire
ator U e = exp(iϕ̂) acts as a discrete translation operator
real axis with the operators x̂R and p̂R that apply on the
of angular momentum, with which it obeys the commu-
half-line, that followed from the Stone-von-Neumann the-
tation relations
orem. Also in that case the mathematical unitary equiv-
alence relates two physically very different situations.
[L̂, U
e] = U
e , e † ] = −U
[L̂, U e† . (73)
Defining angular eigenstates |ϕi by ϕ̂|ϕi = ϕ|ϕi, the C. Physical significance of θ

operator Uα = exp(iL̂α), with α ∈ [−π, π), acts as
For circular motion, the parameter θ is well-known to
Uα |ϕi = exp(iθ)|ϕ − α + 2πi, for − 2π ≤ ϕ − α < −π, represent a magnetic flux that threads the circle, which
Uα |ϕi = |ϕ − αi, for − π ≤ ϕ − α < π, affects a charged particle moving around the circle via
an Aharonov-Bohm phase. First of all, since the angles
Uα |ϕi = exp(−iθ)|ϕ − α − 2πi, for π ≤ ϕ − α < 2π.(74)
ϕ = ±π parametrize one and the same point on the circle,
in the absence of magnetic flux the value of the wave func-
As a result of eq.(73), one then obtains the Weyl group tion should be unique at that point, i.e. Ψ(π) = Ψ(−π),
relation such that θ = 0. In the presence of a magnetic flux Φ,
the angular momentum operator takes the form
Uα U
e = exp(iα)U
e Uα . (75)
eΦ θ
L̂0 = −i∂ϕ + = −i∂ϕ + , θ = eΦ . (78)
The action of the operators Uα and U e is illustrated in 2π 2π
Fig.4. Angular momentum is quantized in integer units Here −e is the charge of the particle and (along with
θ
(shifted by 2π ), and the corresponding discrete transla- ~) we have put the velocity of light to 1. One can now
tion symmetry is generated by U e . The continuous an- perform the unitary transformation
gular rotations described by Uα = exp(iL̂α) act like the
periodic “conveyor belt” associated with Va = exp(ip̂R a) θ
U (θ) = exp i ϕ̂ ,
for the interval that is illustrated in Fig.3. 2π
The similarities between eq.(73) for the circle S 1 = L̂ = U (θ)L̂0 U (θ)† = −i∂ϕ , Ψ(ϕ) = U (θ)Ψ0 (ϕ) ,(79)
[−π, π] and eq.(65) for the interval [0, L] are not acciden-
tal. In fact, the operator pairs L̂, Ve and p̂R , Veπ/L are which implies
mathematically related by a unitary transformation, al-
θ 0 θ
though they are physically quite different. Up to phase Ψ(π) = exp i Ψ (π) = exp i Ψ0 (−π)
factors, the unitary transformation U maps the position 2 2
eigenstates |x, ±i in the interval x ∈ [0, L] to the angular = exp(iθ)Ψ(−π) . (80)
15
After the unitary transformation, the wave function Ψ(ϕ) Under a general gauge transformation ϕ(x) the vector
is no longer single-valued at ϕ = ±π. This is no problem, potential and the wave function transform as
because the unitary transformation corresponds to a non-
periodic gauge transformation. Like the magnetic flux Φ,
ϕ
Ax (x) = Ax (x) − ∂x ϕ(x),ϕ Ψ0 (x) = exp (ieϕ(x)) Ψ0 (x) .
the parameter θ itself is gauge invariant. However, the (87)
unitary transformation U (θ) moves θ from the operator The gauge string connecting the two ends of the interval,
L̂0 to the (no longer strictly periodic) wave function Ψ(ϕ). capped by the values of the wave function at the end
The similarity with the situation in the interval sug- points,
gests that a similar interpretation exists for the corre- Z L
!
sponding parameter θ, which results from S = |0i exp ie dx Ax (x) hL| ,
0
∗ (1 − λ)(1 + λL ) !
exp(iθ) = σσL = . (81) Z L
(1 + λ)(1 − λL ) 0 0 0
hΨ |S|Ψ i = Ψ (0) exp ie ∗
dx Ax (x) Ψ0 (L), (88)
0
As we have seen, the unitary transformation W =
exp(iωσ1 ) of eq.(45) changes the values of the self-adjoint is gauge invariant, i.e. hϕ Ψ0 |ϕ S|ϕ Ψ0 i = hΨ0 |S|Ψ0 i. In
extension parameters, such that σ 0 = exp(−2iω)σ, σL 0
= cases where such a gauge string stretches through the
exp(−2iω)σL , but leaves θ invariant (cf. eq.(63)). Still, interval, the parameter θ appears in the corresponding
we can perform another x-dependent unitary transforma- momentum operator.
tion,

θ θ D. Momentum measurements in [0, L]
W (θ) = exp i σ1 x̂ = exp i x̂R , (82)
2L 2L
which implies σ 0 = σ and σL0
= exp(−iθ)σL such that Let us now consider momentum measurements in the
0 0 ∗
σ σL = 1. The transformation W (θ) again represents a interval [0, L]. For simplicity, we consider a Hamiltonian
gauge transformation in Hilbert space, which now moves without a potential (i.e. V (x) = 0). First, we investigate
θ from the boundary conditions on the wave function to Neumann boundary conditions, which are characterized
the momentum operator by γ = 0 at both ends of the interval. In the limit µ → ∞,
the finite-energy eigenstates of Ĥ(µ) then take the form
θ
p̂0R = W (θ)† p̂R W (θ) = −iσ1 ∂x + . (83) π 2 l2
2L Ĥ(µ)ψl (x) = El ψl (x) , El = , l ∈ N≥0 ,
2mL2
When we perform the unitary transformation on the 1 1
kinetic energy T̂ we obtain ψ0 (x) = √ ,
2L 1

p̂02 1 cos(πlx/L)
T̂ 0 = W (θ)† T̂ W (θ) = 1, ψl>0 (x) = √
L cos(πlx/L)
. (89)
2m
θ
p̂0 = −i∂x + = p̂ + eAx , θ = 2eAx L . (84) Again for simplicity, we choose λL = λ which implies θ =
2L 0, such that the corresponding momentum eigenvalues
In this case, θ manifests itself as a vector potential Ax , and eigenfunctions are
RL
or more precisely as its line integral 0 dxAx = Ax L πn 1−λ
that connects the two boundaries. The original Robin p̂R φk (x) = kφk (x), k = , n ∈ Z, σ = ∈ U (1) ,
L 1+λ
boundary condition, γΨ(0)−∂x Ψ(0) = 0, of eq.(3), along
with its counterpart at the other end of the interval, 1 exp(ikx) + σ exp(−ikx)
φk (x) = √ . (90)
γL Ψ(L) + ∂x Ψ(L) = 0, must also be transformed ac- 2 L exp(ikx) − σ exp(−ikx)
cordingly, and one obtains [35] First of all, one obtains hψl |p̂R |ψl i = 0. Irrespec-
0 0 0
γΨ (0) − Dx Ψ (0) = 0 , γL Ψ (L) + Dx Ψ (L) = 0 ,0 tive of the value of λ, when one projects them onto the
finite-energy sector, the momentum eigenstates are just
Dx = ∂x + ieAx . (85)
φ+ + 1
k,e (x) = φk,o (x) = 2 L exp(ikx). In the ground state
√
The covariant derivative Dx also enters the conserved the probability to measure the momentum value k = 0
probability current is |hφ0 |ψ0 i|2 = 21 . Similarly, the probability to measure
k = πn L with n 6= 0 is
1
j 0 (x) = (Ψ0 (x)∗ Dx Ψ0 (x) − Dx Ψ0 (x)∗ Ψ0 (x)) . (86)
2mi 2
|hφk |ψ0 i|2 = , (91)
π 2 n2
As a result, the boundary conditions of eq.(85) still guar-
antee that no probability leaks out of the interval, i.e. for odd n and zero otherwise. When one measures the
j 0 (0) = j 0 (L) = 0. momentum in an energy eigenstate ψl (x) with l > 0, one
16
obtains k = ± πl 1
L each with probability 4 . The probabil- operator p̂R seem to commute (which would imply a van-
πn
ity to measure k = L for n 6= ±l is then given by ishing momentum uncertainty), the domain incompati-
bility leads to a completely different result.
4n2 Let us now consider the limit L → ∞ in which we
|hφk |ψl i|2 = , (92)
π 2 (l2 − n2 )2 expect to recover the results for the half-line. For γ = 0,
according to eq.(6), R(p) = 1. Then, using eq.(19), for
if (−1)n = −(−1)l and zero otherwise. Indeed the various k 6= ±p one obtains
probabilities, which are illustrated in Fig.5, are correctly
normalized because
4k 2 p2
1 X 2 |hk|ψE i|2 = , E= . (95)
+ = 1 , for l = 0 , (p2 − k 2 )2 2m
2 π n2
2
n∈Z,n odd
1 1 X 4n2 For finite L we identify k = πnL and p =

πl
L, such that,
+ + = 1, for odd l > 0, using eq.(92) for n 6= ±l, one gets
4 4 π (l − n2 )2
2 2
n∈Z,n even
1 1 4n2 4n2 1 4k 2
|hφk |ψl i|2 =
X
+ + = 1, for even l > 0.(93) = 2 2
4 4 π 2 (l2 − n2 )2 π 2 (l2 2
−n ) 2 L (p − k 2 )2
n∈Z,n odd
1
= |hk|ψE i|2 . (96)
L2
0.25
The factor L12 is due to the fact that |ki and |ΨE i are
normalized to δ-functions, while |φk i and |ψl i are normal-
0.20
ized to 1. In any case, this implies that (with Neumann
boundary conditions) in the stationary scattering state
0.15 ψE (x) = exp(−ipx)+exp(ipx) the probability to measure
a momentum value k = p or k = −p is 41 in both cases. In
0.10
the remaining half of the cases the momentum measure-
ment returns a result k 6= ±p with a divergent momen-
tum uncertainty. This somewhat counter-intuitive result
0.05
is again due to domain incompatibilities.
Finally, let us also discuss the standard textbook case
0.00
-10 -9 -8 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9 10
of Dirichlet boundary conditions which are characterized
by γ = ∞. The spectrum of finite-energy states then
FIG. 5. Probability to measure the momentum k = L π
n in takes the form
the energy eigenstate ψl (x) with l = 7 for Neumann boundary
conditions, as a function of n ∈ {−10, . . . , 10}.
π 2 l2
Ĥ(µ)ψl (x) = El ψl (x) , El = 2
, l ∈ N>0 ,
2mL
1 sin(πlx/L)
Let us also consider the momentum uncertainty ψl (x) = √ . (97)
(∆pR )2 = hp̂2R i − hp̂R i2 in the energy eigenstate ψl (x). L sin(πlx/L)
Besides hp̂R i = 0 one obtains
In this case, the probability to measure the momentum
π2 X 2 k = p or k = −p is again 41 in both cases, and the prob-
hp̂2R i = 2 → ∞ , for l = 0 ,
L π2 ability to measure k 6= ±p is
n∈Z,n odd
 
π 2
l2 (−l)2 4n4 4l2
|hφk |ψl i|2 = , (98)
X
hp̂2R i = 2  + + 
π 2 (l2 − n2 )2
L 4 4 π 2 (l2 − n2 )2
n∈Z,n even
→ ∞ , for odd l > 0 , if (−1)n = −(−1)l and zero otherwise. Again, the prob-
abilities, which are illustrated in Fig.6, are correctly nor-
 
2
π l2 (−l)2 X 4n4 malized because
hp̂2R i = 2  + + 
L 4 4 π 2 (l2 − n2 )2
n∈Z,n odd
1 1 X 4l2
→ ∞ , for even l > 0 . (94) + + = 1, for odd l > 0,
4 4 π 2 (l2 − n2 )2
n∈Z,n even
Hence, for Neumann boundary conditions the momen-
tum uncertainty diverges for any energy eigenstate. This 1 1 X 4l2
+ + = 1, for even l > 0.(99)
is a consequence of the domain incompatibility D(T̂ ) 6= 4 4 π 2 (l2 − n2 )2
n∈Z,n odd
D(p̂R ). Although at a formal level of differential expres-
sions the kinetic energy operator T̂ and the momentum In this case one obtains
17
0.25 The additional continuum states correspond to lattice

states with energies at the cut-off scale. In the con-
0.20
tinuum limit, these states are removed from the phys-
ical energy spectrum. The key insight underlying the
new concept is that these states, although they are ulti-
0.15 mately removed to infinite energy, must be kept in the
physical description in order to facilitate the construc-
0.10 tion of a self-adjoint momentum operator. Interestingly,
although the resulting momentum operator for the half-
line is endowed with a self-adjoint extension parameter λ
0.05
(associated with the origin) and is thus not unique, the
results of momentum measurements performed on finite-
0.00
-10 -9 -8 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9 10
energy states are independent of this parameter. In an
interval, the momentum operator is characterized by two
FIG. 6. Probability to measure the momentum k = L π
n in self-adjoint extension parameters λ and λL (associated
the energy eigenstate ψl (x) with l = 7 for Dirichlet boundary with the two boundary points). In that case, the value
π θ
conditions, as a function of n ∈ {−10, . . . , 10}. of the quantized momentum, k = L (n + 2π ), depends
on the particular combination θ of the two self-adjoint
  extension parameters.
π l2
(−l) 2 2 X 4l n2 2 Based upon the new concept for the momentum op-
hp̂2R i = + +  erator, canonical quantization becomes applicable both
L2 4 4 π 2 (l2 − n2 )2
n∈Z,n even to the half-line and to an interval. However, due to the
π l 2 2 existence of sharp boundaries, self-adjoint extension pa-
= , for odd l > 0 , rameters enter the description and thus lead to physically
L2  in-equivalent quantum variants of the same classical sys-
π 2 l2 (−l)2 X 4l2 n2 tem. This goes beyond the usual operator ordering am-
hp̂2R i = 2  + + 
biguities. In particular, different operators, like the mo-
L 4 4 π 2 (l2 − n2 )2
n∈Z,n odd
mentum p̂R and the Hamiltonian Ĥ(µ), act in different
2 2
π l domains, D(p̂R ) 6= D(Ĥ(µ)), of the Hilbert space. As a
= , for even l > 0 , (100)
L2 consequence, the commutation relations that result from
classical Poisson bracket relations are just formal equa-
which implies (∆pR )2 = 2mEl . Hence, for Dirich-
tions relating differential expressions. Understanding the
let boundary conditions the momentum uncertainty in
true nature of the relations between the various operators
an energy eigenstate is finite. This is because, in this
requires a careful analysis of the corresponding operator
case, D(T̂ ) ⊂ D(p̂R ). Again using eq.(6), one now gets
domains.
R(p) = −1, and following eq.(19), for k 6= ±p one obtains
This is an inevitable consequence of the low-energy
4p2 continuum description of a system with sharp impene-
|hk|ψE i|2 = = L2 |hφk |ψl i|2 . (101) trable boundaries, which are necessarily ultraviolet sen-
(p2 − k 2 )2
sitive. Working explicitly with an ultraviolet lattice cut-
For Dirichlet boundary conditions the stationary scat- off (representing the shortest physically relevant distance
tering state is ψE (x) = exp(−ipx) − exp(ipx). Then the scale) is straightforward and might even be quite phys-
probability to measure momentum k = p or k = −p is ical, but is not very transparent. A transparent ef-
again 14 in both cases. In the remaining half of the cases fective low-energy continuum description, as it is com-
the momentum measurement results in k 6= ±p, now with pletely common in quantum mechanics, necessarily uses
a finite momentum uncertainty. an infinite-dimensional Hilbert space. As we have seen,
on the half-line or in an interval this requires a careful ap-
plication of the canonical quantization procedure. Famil-
V. CONCLUSIONS iarizing ourselves with the elegant mathematical frame-
work originally established by von Neumann is very well
We have introduced a new concept for a self-adjoint worth the effort in order to gain a more complete physical
quantum mechanical momentum operator for an inter- understanding of these “simple” quantum systems.
val [0, L] and for the half-line R≥0 . The new concept It should be mentioned that most of the somewhat
arises naturally in the continuum limit of the lattice- subtle Hilbert space and operator domain issues, associ-
regularized problem. On the lattice one distinguishes ated with the canonical quantization procedure that we
even and odd lattice points. In the continuum limit, this carried out above, arose only because we decided to con-
naturally leads to a two-component wave function, which struct the momentum operator in addition to just the
is associated with a doubling of the Hilbert space from Hamiltonian. To a physicist who favors Everett’s many-
L2 ([0, L]) to L2 ([0, L]2 ) and from L2 (R≥0 ) to L2 (R2≥0 ). worlds interpretation of quantum mechanics [47], this
18
may seem unnecessary, because any measurement pro- segment, such that x = (n − 12 )a, n ∈ {1, 2, . . . , N }.
cess is then incorporated in the global Hamiltonian that The Hilbert space then becomes N -dimensional and self-
governs the time-evolution of the wave function of the en- adjointness becomes indistinguishable from Hermiticity.
tire universe. This wave function includes the quantum As a result, there are no longer any domain issues, since
system under study, any device that registers measure- every operator can act in the entire finite-dimensional
ment results, as well as the conscious observer who reads Hilbert space.
off those results and uses them to draw conclusions about
how the quantum system works. Still, when engaging in
an actual study of an isolated quantum system (rather
than of the universe as a whole), even a hard-line “Ev- a
a/2
erettian” would probably prefer to work with the much
FIG. 7. Lattice with N = 9 points in the interval [0, L].
more tractable canonical quantization procedure applied
above.
The new concept of momentum in a space with bound- The lattice momentum operator is represented by for-
aries has potentially far reaching consequences. Its gen- ward and backward discretized derivatives.
eralization to higher dimensions is straightforward and 
−1 1 0 . . . 0 0 0

was sketched in [9]. A natural next step, which is cur-  0 −1 1 . . . 0 0 0 
rently under investigation, concerns the generalization  0 0 −1 . . . 0 0 0 
 
and physical interpretation of the Heisenberg uncertainty i .

. . ... . . . 

relation that was derived for a finite volume with sharp p̂F = −   ,
a . . . ... . . . 
boundaries in [35]. This can be applied, for example, to  0 0 0 . . . −1 1 0 
 
optical box traps [1] and quantum dots [2], which may  0 0 0 . . . 0 −1 1 
make the new momentum concept experimentally acces- 0 0 0 . . . 0 0 λL
sible. In this context, it is also interesting to construct a
−λ 0 0 . . . 0 0 0
 
momentum measurement device, at least at a theoretical
 −1 1 0 . . . 0 0 0 
level, for example, along the lines originally introduced
 0 −1 1 . . . 0 0 0 
 
by von Neumann [27]. This implies to couple the quan- i . . . ... . . . 
 
tum particle to another quantum system that serves as a p̂B = −   . (A1)
measurement device, whose pointer can then be read out a . . . ... . . . 
 0 0 0 ... 1 0 0 
at the classical level. Time-of-flight momentum measure- 
 0 0 0 . . . −1 1 0 

ments of this kind have been discussed, for example, in
0 0 0 . . . 0 −1 1
[48]. Further applications, again along the lines of [35],
suggest themselves in the context of relativistic fermions, On the lattice, the parameters λ, λL ∈ iR are directly
for example, to the phenomenological MIT bag model incorporated in the corresponding matrices. In the con-
[3–5], or to domain wall fermions residing in an interval tinuum limit a → 0 they turn into self-adjoint extension
of extra-dimensional space [6, 7]. Canonical quantiza- parameters. Neither p̂F nor p̂B are Hermitean matrices.
tion (which was thought to be inapplicable because the It is natural to construct the following combinations
standard momentum operator is not self-adjoint) can be
1
applied in all these situations based on the new concept p̂R = (p̂F + p̂†F + p̂B + p̂†B )
of momentum in a space with sharp boundaries. 4
−λ 1 0 . . . 0 0 0
 
 −1 0 1 . . . 0 0 0 
 0 −1 0 . . . 0 0 0 
 
VI. ACKNOWLEDGMENTS
i  .

. . ... . . . 

=−   ,
2a  . . . ... . . . 
We thank Matthias Blau for illuminating discussions.  0 0 0 ... 0 1 0 
 
 0 0 0 . . . −1 0 1 
0 0 0 . . . 0 −1 λL
Appendix A: Lattice regularization of the
1
momentum operator ip̂I = (p̂F − p̂†F + p̂B − p̂†B )
4
i
In order to circumvent the subtleties associated with = diag(1, 0, 0, . . . , 0, 0, −1) . (A2)
Hermiticity versus self-adjointness, which arise because 2a
the Hilbert space is infinite-dimensional, in [9] we have The resulting momentum operator p̂R + ip̂I is not Her-
investigated the problem on the interval in an ultraviolet mitean, but has a Hermitean component p̂R and an anti-
lattice regularization. As illustrated in Fig.7, the interval Hermitean component ip̂I , which is diagonal in the posi-
[0, L] is then divided into N = L/a segments of size a (not tion basis. The Hermitean component p̂R corresponds to
to be confused with the self-adjoint extension parameter a symmetrized forward-backward next-to-nearest neigh-
of eq.(38)), with a lattice point in the middle of each bor derivative that extends over two lattice spacings.
19
The lattice eigenvalue problem of p̂R is given by This relation reduces to eq.(61) in the continuum limit.
1
p̂R φk,x = sin(ka)φk,x , (A3)
a
and the momentum quantization condition takes the
form
(1 − λ exp(ika))(1 + λL exp(ika))
exp(2ikL) = . (A4)
(exp(ika) + λ)(exp(ika) − λL )
[1] A. L. Gaunt, T. F. Schmidutz, I. Gotlibovych, R. P. [27] J. von Neumann, “Mathematische Grundlagen der Quan-
Smith, and Z. Hadzibabic, Phys. Rev. Lett. 110 (2013) tenmechanik”, Springer, Berlin (1932).
200406. [28] M. Reed and B. Simon, “Methods of Modern Mathe-
[2] P. Harrison, Quantum wells, wires and dots, John Wiley matical Physics II, Fourier Analysis, Self-Adjointness”,
and Sons Ltd. (2005). Academic Press Inc., New York (1975).
[3] A. Chodos, R. L. Jaffe, K. Johnson, C. B. Thorn, and V. [29] F. Gieres, Rep. Progr. Phys. 63 (2000) 1893.
F. Weisskopf, Phys. Rev. D9 (1974) 3471. [30] T. Jurić, arXiv:2103.01080.
[4] A. Chodos, R. L. Jaffe, K. Johnson, and C. B. Thorn, [31] R. Balian and C. Bloch, Ann. Phys. 60 (1970) 401.
Phys. Rev. D10 (1974) 2599. [32] T. E. Clark, R. Menioff, and D. H. Sharp, Phys. Rev.
[5] P. Hasenfratz and J. Kuti, Phys. Rept. 40 (1978) 75. D22 (1980) 3012.
[6] D. B. Kaplan, Phys. Lett. B288 (1992) 342. [33] E. Farhi and S. Gutmann, Int. J. Mod. Phys. A5 (1990)
[7] Y. Shamir, Nucl. Phys. B406 (1993) 90. 3029.
[8] G. Bonneau, J. Faraut, and G. Valent, Am. J. Phys. 69 [34] M. Carreau, E. Farhi, and S. Gutmann, Phys. Rev. D42
(2001) 322. (1990) 1194.
[9] M. H. Al-Hashimi and U.-J. Wiese, arXiv:2012.09596. [35] M. H. Al-Hashimi and U.-J. Wiese, Ann. Phys. 327
[10] H. B. Nielsen and M. Ninomiya, Phys. Lett. B105 (1981) (2012) 1.
219; [36] J. Twamley, G. J. Milburn, New J. Phys. 8 (2006) 328.
[11] H. B. Nielsen and M. Ninomiya, Nucl. Phys. B185 (1981) [37] J. R. Klauder, J. Phys. A: Math. Theor. 45 (2012)
20. 285304.
[12] D. Friedan, Commun. Math. Phys. 85(1982) 481. [38] H. Bergeron and J.-P. Gazeau, Ann. Phys. 344 (2014) 43.
[13] E. B. Davies and J. T. Lewis, Commun. Math. Phys. 17 [39] C. R. Almeida, H. Bergeron, J.-P. Gazeau, and A. C.
(1970) 239. Scardua, Ann. Phys. 392 (2018) 206.
[14] K. Kraus, Ann. Phys. 64 (1971) 311. [40] L. Gouba, arXiv:2005.08696.
[15] A. S. Holevo, Rep. Math. Phys. 13 (1978) 379. [41] C. Cohen-Tannoudji, B. Dui, and F. Laloë, Quantum
[16] A. S. Holevo, Rep. Math. Phys. 16 (1979) 385. Mechanics I, Wiley, New York (1977).
[17] E. B. Davies, “Quantum Dynamics of Open Systems”, [42] P. Garbaczewski and W. Karwowski, Am. J. Phys. 72
Academic Press, New York (1976). (2004) 924.
[18] A. S. Holevo, “Statistical Structure of Quantum Theory”, [43] M. H. Stone (1930), “Linear Transformations in Hilbert
Lect. Not. Phys. 61, Springer, Berlin (2001). Space. III. Operational Methods and Group Theory”,
[19] K. Kraus, “States, Effects, and Operations: Fundamental Proceedings of the National Academy of Sciences of the
Notions of Quantum Theory”, Springer, Berlin (1983). United States of America, National Academy of Sciences,
[20] A. Peres, Found. Phys. 20 (1990) 1441. 16 (1930) 172.
[21] M. A. Naimark, Iza. Akad. Nauk USSR, Ser. Mat. 4, 53 [44] J. von Neumann, “Die Eindeutigkeit der Schrödinger-
(1940) 277. schen Operatoren”, Mathematische Annalen, Springer
[22] M. A. Naimark, C. R. Acad. Sci. URSS 41 (1943) 359. Berlin / Heidelberg, 104 (1931) 570.
[23] N. I. Akhiezer and I. M. Glazman, “Theory of Linear [45] J. von Neumann, “Über einen Satz von Herrn M. H.
Operators in Hilbert Space”, Ungar, New York (1963). Stone”, Annals of Mathematics, 33 (1932) 567.
[24] C. W. Helstrom, Int. J. Theor. Phys. 8 (1973) 361. [46] M. H. Stone, “On one-parameter unitary groups in
[25] F. Belgiorno and F. Gallone, J. Math. Phys. 50 (2009) Hilbert Space”, Annals of Mathematics, 33 (1932) 643.
022101. [47] H. Everett, Rev. Mod. Phys. 29 (1957) 454.
[26] Y. Shikano and A. Hosoya, J. Math. Phys. 49 (2008) [48] F. Di Pumpo and M. Freyberger, Eur. Phys. J. D73
052104. (2019) 163.

Canonical Quantization On The Half-Line and in An Interval Based Upon A New Concept For The Momentum in A Space With Boundaries

Uploaded by

Copyright:

Available Formats

Canonical Quantization On The Half-Line and in An Interval Based Upon A New Concept For The Momentum in A Space With Boundaries

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Canonical Quantization On The Half-Line and in An Interval Based Upon A New Concept For The Momentum in A Space With Boundaries

Uploaded by

Copyright:

Available Formats

Canonical quantization on the half-line and in an interval based upon

a new concept for the momentum in a space with boundaries

I. INTRODUCTION blocks of higher-dimensional regions of space with sharp

1 The position operator x̂ = i∂p generates infinitesimal

|x − ai |xi γ → 0− becomes proportional to the step function θ(x),

Since the standard momentum operator p̂ = −i∂x is 1

As expected, the resulting momentum expectation value

• Realize the commutation relations by differential

eigenstates |ϕi on the circle ϕ ∈ [−π, π]

Defining angular eigenstates |ϕi by ϕ̂|ϕi = ϕ|ϕi, the C. Physical significance of θ

1 1 X 4n2 For finite L we identify k = πnL and p =

0.25 The additional continuum states correspond to lattice

You might also like